The AI visibility test reveals whether or not your product is recommended for relevant AI-generated responses about your category.
It answers questions like:
- When someone asks ChatGPT, Perplexity, or Google AI about our category, does our product show up?
- What topics is our product recommended for, and where are we lacking?
- Is our current health website and content strategy set up to help us succeed in AI search?
This guide shows you how to use Semrush to answer these questions with an AI visibility test. You will measure where the product is appearing on all AI platforms, what content is being cited, and where there are topic gaps.
Don’t have Semrush yet? Try our free AI visibility trial to get started without commitment.
What to Include in AI Visualization Research
Focus your research on these six characteristics:
- Overall Visibility and Voice Sharing – How often your product appears compared to competitors across all AI platforms
- Product Characteristics – Often your brand name appears in AI generated responses
- Website Citations – What specific pages does the AI pull from when making responses
- Subject Spaces and Emotions – Topics where competitors are well or clearly designed
- Content Opportunities – Questions the AI already answers, but without specifying your type
- Technical Readiness – That AI searchers can access and interpret your content
These metrics establish clear benchmarks and point to the most important next steps
Step 1: Establish Your Foundation for AI Visibility
Start by measuring how your product currently appears in the AI-generated responses you’re using Overview in the Semrush AI Visibility toolkit.
This dashboard shows mention counts, pages cited, and distribution across all AI platforms (ChatGPT, Google AI Overview, Perplexity, Gemini).
Note: Write down your AI score for visibility, mentions, and pages cited. These high-level metrics can be used to benchmark competitors to see if AI systems refer to your brand name or cite your content as a source more than your competitors.
On Allbirds, we see 7.4K mentions and 581 citation pages.
This foundation can now be compared to other competing brands to see if you have a visibility problem that needs to be addressed.
Step 2: Find Content Worth Citing
Next, check which pages the AI systems cite when generating answers. Navigate to “Pages Cited” section on Visibility Overview.
This shows which URLs from the domain are being referred to in the AI responses and how many references from the Semrush database each page has now been cited.
Tutorial and comparison pages get more citations than product or category pages.
For Allbirds, their homepage is the most cited (90 mentions), but specific content (such as hiking shoe guides, walking vs. running FAQs, and shoe care) is also frequently cited.
Expand the page to see the information and whether the brand name is “Mentioned” or “Missed” within those responses that cite the page.
-
Quoted and stated: this is a win for visibility
- Quoted but not mentioned: this would be an opportunity to update the page and mention it in this material (as long as other species are mentioned)
Compare AI Indexed Pages and SEO Top Pages
You can use the Top Pages report to compare these with the pages that attract the most SEO traffic. Note in the case of Allbirds, organic traffic is mostly driven by the collection and category pages.
These pages already rank well in regular searches. Now develop them for AI retrieval—add a clear structure, use question-and-answer formats, and concise summaries that AI systems can extract and say.
Step 3: Get Titles and Alerts When Your Name Doesn’t Exist
Topic spaces reveal when the AI recommends your competitors and leaves your product in response.
In the visibility overview, scroll to “Topics and Sources” and choose “Title Possibilities.”
Each topic shows where AI-generated answers exist and attract demand. Expanding the topic reveals specific instructions that generate those responses.
For Allbirds, high-volume topics like “Running Shoes” and “Diabetic and Neuropathy Shoes” show 3 competitors mentioned—but not Allbirds. Each topic can be expanded to reveal specific instructions that create those responses.
Action: Note the missed instructions for identifying questions your existing content doesn’t answer, and then create content that answers those questions clearly and directly.
Step 4. Analyze Brand Sentiment and AI Narrative
Unlike search results, AI does not rank products by their URL. However, it can describe a product as an option that has pros and cons alongside other options.
use the Product Performance reports to review sentiment across orders and see which brands are framed as positive, neutral, or negative.
In this summary, Allbirds is often framed correctly (88%) and captures a larger share of voice (14%) in AI responses compared to the competition.
You can navigate to the Key Emotion Drivers table in the Insights Report to review the product’s strengths and areas for improvement—perfect for incorporating into your testing.
Some of the widgets in the Product Performance section go deeper into exploring the factors that contribute to a good product experience and the areas that should be improved in AI responses.
So be sure to check all those reports, or generate a PDF from this tool with one click.
Step 5. Find Influential Off-Site Sources Shaping AI Answers
AI-generated answers often cite third-party UGC platforms such as Reddit, YouTube, Quora, LinkedIn, and Medium. After Visibility Overview, check “Sources Cited” to see which external domains are indexed the most across your main topics.
The goal is to assess how influential these platforms are in your niche. When sites like YouTube and Reddit pop up again and again, they shape how AI systems define a category and what types are identified.
You can use this data to decide if it’s worth investing in growing your brand on YouTube or managing your reputation on Reddit.
Step 6: Check for Site-Level Issues Affecting AI Access
Use it Site Inspection identifying technical issues that make pages difficult for AI systems to interpret and cite.
Common issues affecting AI visibility:
- Weak internal content: Links without descriptive anchor text slow down the way pages are indexed clearly in AI systems
- Pages divided by structure: Pages with only one inbound link are less likely to be indexed regularly
- The AI directive file is missing: There is no llms.txt that does not block access, but removes a clear signal of how the AI should interpret the site’s content.
These issues don’t prevent AI from crawling, but they do explain why citations appear on a small set of clear pages. Automating makes many pages easier for AI systems to use as sources.
How to Bring Your Research to AI Visibility
Focus on presenting measurable and actionable findings.
The most useful tests reveal patterns: when AI includes or excludes a brand, what types of content AI relies on, and how similar information can be handled across categories.
Here are three delivery options:
- Document Illustrated with Screenshots: Combine Semrush’s main screenshots with brief comments on what each report shows and what it means. It’s quick to produce and easy to share via email.
- Standard PDFs with Semrush My Reports: Use Semrush’s My Reports templates to aggregate metrics in one branded format. By connecting Google Analytics, this can include AI referral traffic analysis widgets. These are great for frequent client reports.
- Video walkthrough: Pair a written assessment with a Loom recording or live presentation to explain how the findings connect, highlight important points, and clarify trade-offs. It is best if the preparation is needed before the action.
Choose a format based on how much context your recipient needs to decide what to do next.
What to Do After Your AI Visibility Test
Once you have research, you can decide what to do next. The roadmap for improving AI visibility should be different for each business and situation, but here is a suggested starting point.
Fast (Week 1): Fix Technical Issues Limiting AI Access
Resolve issues that affect whether AI systems can access, interpret, or describe your content. This modification affects the entire site and sets the foundation for everything that follows.
Near Term (Weeks 2–4): Develop Quotation Readiness
Focus on developing unpublished content. Make existing pages clearer, more complete, and easier for AI systems to reuse. After a month, you can do your research again to see how much progress has been made.
Medium Term (2–3 Months): Extend Subject Coverage
Address the topic gaps by creating new content in those areas where your product is lacking.
Turn AI Visualization Test Findings Into Key Values
As AI systems increasingly shape how options are compared and recommended, many brands still lack visibility into why their brand appears in some responses and not others.
With Semrush, AI search patterns become visible signals that you can review, prioritize, and revisit over time. Get started with a Semrush AI visibility test before your competitors do and stay ahead of your market.
Banner: Start Your AI Visibility Audit with Semrush
